Microsoft Stops Paying OpenAI Revenue Share; Foreign Acquisition of Manus Blocked

Introduction

On April 27, 2025, several major stories broke simultaneously: the profit-sharing arrangement between Microsoft and OpenAI reached a major turning point, China's NDRC issued a prohibition on a foreign acquisition of the Manus project, the Enterprise State-Owned Assets Law launched its first revision since taking effect nearly 17 years ago, and China's domestic open-source large language model ecosystem hit an inspiring milestone. Spanning commercial rivalries and policy regulation, these events vividly outline the latest trajectory of global AI industry competition.

Microsoft Ends Revenue Sharing With OpenAI, Partnership Model Shifts

Microsoft officially announced it will no longer pay revenue share to OpenAI. The decision marks the formal end of the long-standing "investment-for-revenue-share" partnership model between the two AI giants.

Previously, Microsoft had invested a cumulative total of over $13 billion in OpenAI and deeply integrated OpenAI's GPT-series models into core product lines including Azure cloud services and Microsoft 365 Copilot. Under their agreement, Microsoft was required to return a certain percentage of revenue from related AI products to OpenAI. As OpenAI proceeds with its transition to a for-profit company and Microsoft continues to strengthen its own AI capabilities, the commercial relationship between the two is being redefined.

Industry analysts believe Microsoft's move may carry several signals: first, Microsoft's growing self-sufficiency in AI infrastructure and model capabilities is reducing its technical dependence on OpenAI; second, as OpenAI accelerates commercialization, the original investment-and-share framework can no longer accommodate either party's strategic expansion needs; and third, amid white-hot AI industry competition, Microsoft needs a more flexible cost structure to counter pressure from rivals such as Google and Amazon.

NDRC Blocks Foreign Acquisition of Manus Project

China's NDRC issued a prohibition on foreign investment in the acquisition of the Manus project in accordance with applicable laws and regulations. Manus had previously attracted significant attention for its technological breakthroughs in the AI Agent domain and was regarded as one of China's flagship projects in the AI application layer.

The prohibition reflects China's cautious stance toward foreign mergers and acquisitions in critical technology sectors. Against the backdrop of an increasingly complex global AI competition landscape, protecting domestic core AI technology assets and preventing the outflow of key technologies have become important considerations in policymaking worldwide. The decision also resonates with the broader environment of escalating U.S.-China technology rivalry.

Notably, the timing of this decision coincides with the first revision of the Enterprise State-Owned Assets Law since its implementation nearly 17 years ago. The revision will further refine the state-owned asset supervision system, strengthen regulation of major business activities such as investment and M&A by state-owned enterprises, and provide more robust legal safeguards for the preservation and appreciation of state-owned assets in the digital economy era.

Domestic Open-Source LLM Downloads Surpass 10 Billion

On the industry development front, domestically developed open-source large language models have recently undergone intensive iterative upgrades, with cumulative downloads surpassing the 10 billion mark. From Alibaba's Qwen series to DeepSeek and Zhipu GLM, Chinese open-source LLMs are gaining increasingly broad recognition and adoption across global developer communities.

This figure not only reflects the rapid improvement in the technical capabilities of domestic LLMs but also demonstrates that China's AI open-source ecosystem is forming a virtuous cycle — open-source models drive developer community growth, and community feedback in turn fuels model iteration and optimization. In the global AI open-source competition, China's presence can no longer be overlooked.

Other Noteworthy Developments

• Strong Industrial Profit Growth: Data from the National Bureau of Statistics shows that from January to March, industrial enterprises above a designated size nationwide achieved total profits of 1.6960 trillion yuan, a year-on-year increase of 1.5%, providing a macroeconomic foundation for sustained investment in emerging industries such as AI.
• Multiple Central Banks Collectively Sell Off Gold: New shifts in global capital flows have emerged, with some funds potentially accelerating into technology and AI sectors.
• New Regulations for Online Public Fundraising Platforms: The rules explicitly prohibit the insertion of commercial advertisements and stipulate retention periods for fundraising information, reflecting the continued deepening of internet platform regulation.
• Mandatory National Standards for Civil Aviation Take Effect: Two new standards will be implemented starting May 1, further raising requirements for digitalized and intelligent management.

Outlook and Assessment

The global AI industry is currently at a critical juncture in the transition from technological breakthroughs to commercial deployment. The adjustment of the Microsoft-OpenAI partnership model signals that profit-distribution mechanisms across the AI value chain will continue to evolve. The blocking of the Manus acquisition highlights that core AI technology has risen to the level of a national strategic resource. And the flourishing development of domestic open-source LLMs has secured greater strategic initiative for China in global AI competition.

It is foreseeable that in the second half of 2025, competition around AI technology sovereignty, open-source ecosystem development, and business model innovation will intensify further. Whether policymakers, corporate decision-makers, or technology developers, all must find their footing in this transformation and seize the window of opportunity.